Journal
of Financial
Economics
32 (1992) 3-21.
North-Holland
Convertible bonds as backdoor equity
financing
Jeremy
C. Stein*
Massachusetts
Insrirure
Received September
of Technology.
Cambridge,
1991, final version
received
.MA
021.59.
March
1992
LISA
This paper argues that corporations
may use convertible bonds as an indirect way to get equity into
their capital structures when adverse-selection
problems make a conventional
stock issue unattractive. Unlike other theories of convertible bond issuance. the model here highlights: 1) the importance
of call provisions
on convertibles
and 2) the significance
of costs of financial distress to the
information
content of a convertible issue.
1. Introduction
Convertible bonds are an important source of financing for many corporations. According to data presented in Essig (1991), more than 10% of all
COMPUSTAT companies had ratios of convertible debt to total debt exceeding
33% during the period 1963-1984. A good deal of research effort has been
devoted to developing pricing models for convertibles,’ as well as to the issues
surrounding corporations’ policies for calling them.2 Somewhat less work has
addressed the fundamental question of why companies issue convertibles in the
first place.
This paper develops a rationale for the use of convertible debt. I argue that
companies may use convertible bonds to get equity into their capital structures
Correspondence
Technology,
to: Jeremy C. Stein, Sloan
50 Memorial Drive, Cambridge,
School of Management,
MA 02139. USA.
Massachusetts
Institute
of
*This research is supported
by a Batterymarch
Fellowship and by the International
Financial
Services Research Center at MIT. I thank Paul Asquith, Kenneth Froot, Steven Kaplan, Wayne
Mikkelson (the referee). Stewart Myers, Richard Ruback (the editor), and David Scharfstein for
helpful comments.
I am especially grateful to Bruce Greenwald
for sharing his insights about the
MCI case with me.
‘See, e.g., Ingersoll
(1977a) and Brennan
‘See Ingersoll (1977b). Mikkelson
Asquith and Mullins (1991). among
0304-405X,‘92f’%05.00
C
and Schwartz
(1981). Harris
others.
1992-Elsevier
Science
(1977, 1980).
and Raviv (1985). Jaffee and Shleifer (1990). and
Publishers
B.V. All rights reserved
6
2.2. Financing
J.C. Stein, Conuerrible bonds as buckdoor equiry financing
instruments
In the first part of the analysis. I focus on three financing options open to
firms at time 0: equity, straight long-term debt that matures at time 2, and
callable convertible debt that also matures at time 2. For the time being,
short-term debt due at time 1 is not considered. The possibility of short-term
debt is addressed in section 2.4 below.
A policy of equity financing can be summarized by the fraction of the firm’s
time 2 gross cash flows apportioned to outside claimants. A policy of long-term
debt financing can be described by the face value of the debt - i.e., the promised
repayment at time 2. It is assumed that long-term debt financing carries with it
the potential for costly financial distress. If the firm’s cash flow at time 2 falls
short of the face value of the debt, a deadweight cost of c is imposed on the
owner-manager. For the sake of simplicity, this cost c is taken to be exogenous,
and might be thought of as representing time and resources devoted to litigation; etc.5 A more general approach,. and one more suitable for non-ownermanaged firms, might involve endogenizing the extent to which financial distress
imposes costs on shareholders, perhaps by appealing to Myers’s (1977) idea of
underinvestment in the face of a debt overhang, or a related concept.
A convertible bond has both a face value and a conversion ratio. It is also
callable at time 1, so that the issuer may attempt to force conversion at this time.
If investors convert, they get the specified number of common shares. If the bond
is not called, they retain a debt contract with the original face value. Clearly, the
issuer will not always be able to successfully force conversion. Investors will
rationally convert only if the stock price at time I is high enough, so that the
conversion value exceeds the call price.
2.3. A separating
equilibrium
with convertible financing
To illustrate the benefits of convertible financing, I first demonstrate that the
availability of convertible bonds makes it possible to sustain a separating
equilibrium in which all types of firms issue fairly priced claims and invest
efficiently. This outcome is not generally possible when long-term debt and
equity are the only financing options. I then argue in section 2.4 (using a slightly
expanded version of the model) that even when we enrich the range of financing
choices to include short-term debt, it is still not generally possible to achieve an
efficient separating equilibrium without the use of convertibles.
The nature of the efficient separating equilibrium is summarized in the
following proposition:
‘A similar formulation
Diamond (1984).
- nonpecuniary
bankruptcy
costs for the owner-manager
- is adopted
by
J.C.
Proposition.
Srrm,
Concvrrible
If costs offinancial
then the following
bonds
us buckdoor
equlr)
7
jnancmg
distress are high enough, so that c > (I - .U,).
is a separating
equilibrium:
issue debt with a face calue of I and incest;
(i) goodjrms
(ii) bad firms issue a fraction I,‘(qXu + (1 - q)X,) of equity and incest;
(iii) medium firms issue a concertible bond and inrest.
The concertible
has a face
calue of F > X L; has a call price K that is gicen b)
XL < K < I; and is concertible
into a fraction
f/(pX,
+ (1 - p)X,)
of the firm’s
equity.
To prove the proposition,
we need to check three pairs of incentive conditions. Given the market beliefs associated with the equilibrium,
we require that:
1) a bad firm does not wish to mimic either a medium or a good firm; 1)
a medium firm does not wish to mimic either a bad or a good firm; and 3) a good
firm does not wish to mimic either a bad or a medium firm.
2.3.1.
The perspectice
of a hadjrm
Much of the insight for the proposition
is revealed by showing that a bad firm
will not mimic a medium firm by issuing a convertible bond. Given the convertible structure
outlined
above, a bad firm issuing a convertible
would face
a probability
I of being unable to force conversion at time 1 - when a bad firm
deteriorates
(which happens with probability
z), the conversion
value of the
bond falls to fXL/(pXH + (1 - p)X,), which is below the call price K. In this
circumstance,
the call provision cannot be used to force conversion.
Consequently,
if a bad firm issues a convertible, there is a probability
z that it
will be left with a debt burden of F.6 This ‘convertible overhang’ in turn leads to
certain financial distress, since a deteriorated
bad firm’s cash flow of XL always
falls short of F. So, the expected costs of financial distress to a bad firm
associated with a convertible
issue are given by zc.
On the other hand, there is some gain to a bad firm in mimicking a medium
firm, in that it sells an overpriced security. It sells a convertible that it knows will
remain a debt claim worth XL with probability
z and that will convert into
equity worth I with probability (1 - z). The true value of this convertible is thus
(1 - z)l + 2X,, but a bad firm is able to raise an amount I with it. Thus the bad
firm has issued a security that is overpriced by an amount z(f - XL).
Taking both factors into consideration,
a bad firm will choose to issue the
convertible only if the overpricing amount exceeds the expected distress costs of
6A deteriorated
unable to use the
exceeds the entire
it to raise enough
bad firm is not only unable to use the call provision to force conversion. but is also
call provision to buy back the convertible for cash. This is because the call price
market value of a deteriorated
bad firm. so that no method of financing will allow
cash at time I to pay the call price.
8
J C. Srein. Conrerrihk
bonds as buckdoor
equity Jinancing
zc. However, given the condition on c stipulated in the proposition,
namely, that
c > (I - XL). this can never happen. Therefore a bad firm will not mimic
a medium firm. The central role of costs of financial distress is apparent here; if
c were smaller, there would indeed be an incentive for a bad firm to mimic
a medium firm, and the conjectured separating equilibrium
would be destroyed.
The argument that a bad firm will also not mimic a good firm is more direct: if
it does so, it issues debt that has a true value of ql + (1 - q)XL. Since the bad
firm raises an amount I, the debt issue is overpriced by (1 - 4)(1 - XL). But
given the stipulated condition
that c > (I - X,), the overpricing
is again less
than the costs of financial distress (1 - q)c. So a bad firm does not mimic a good
firm.
2.3.2.
The perspective of a medium firm
When a medium firm issues a convertible,
it is issuing a security that it views
as fairly priced. In addition, a medium firm bears no expected costs of financial
distress with a convertible,
because it knows that the conversion
value of the
bond will always be equal to I at time 1. Since this conversion value exceeds the
call price K, a medium firm can force conversion with certainty and is never left
with any debt burden.’
In view of this fact, it is obvious that a medium firm will not wish to mimic
a bad firm. Such mimicking would entail the medium firm’s issuing an underpriced (from its perspective)
equity security and gaining nothing in return.
Unlike a bad firm, a medium firm cannot reduce expected distress costs by
switching to equity financing, since it already perceives these costs to be zero.
A medium firm will also not wish to mimic a good firm, for the same reasons
that a bad firm will not wish to mimic a good firm. As before, the expected
distress costs associated with long-term debt outweigh the benefits from selling
an overpriced security. Indeed, the logic is identical to that sketched above for
a bad firm, with 4 replaced everywhere by p.
2.3.3.
The perspectice of a good firm
There is clearly no reason for a good firm to deviate from a policy of
long-term debt financing. A good firm bears no expected costs of distress with
long-term debt. Thus by mimicking a bad or a medium firm, it can only sell what
it sees as an underpriced
security, with no compensating
benefit.
In summary, the logic for a convertible is this: a medium firm does not want to
issue equity because of the negative inference the market would draw. Straight
‘It is easily checked that it is always in a medium firm’s best interest to use the call provision to
force conversion at time 1. This would be true even without costs of financial distress - by forcing
conversion a medium firm extinguishes
an otherwise valuable option held by investors.
J.C. Stein. Conrerttble bonds us bockdoor equity financing
9
debt is also unattractive,
because of the costs of financial distress. A convertible
allows a medium firm to get equity into its capital structure, while at the same
time conveying
a more positive message to the market: a convertible
issue
cannot be coming from a bad firm, since a bad firm knows that its stock price
may not be high enough at time 1 to realize conversion.
In this framework, the use of convertible
securities has positive efficiency
implications.
In the separating equilibrium
described above, all firms invest, and
no costs of distress are actually borne in equilibrium.
If convertibles
were not
available, this would not generally be the case. Rather, we would be back in
a situation very similar to that described in the original Myers and Majluf (1984)
paper, and might well have some types of firms failing to invest. With only two
modes of financing (equity and long-term debt) available, any equilibrium
that
had all types of firms investing would have to involve some pooling in terms of
financing choices. But if the NPV of the investment
is relatively small, such
pooling equilibria cannot be sustained - the higher-quality
firm in the pool will
prefer to pass up the investment
instead of issuing what it perceives to be an
underpriced
security.*
2.4.
Why don’t medium firms
simply postpone
equity issues until time I?
We have thus far not explored the possibility that firms might issue short-term
debt and then refinance it at time 1 with an equity issue. If this option were
introduced
into the current set-up, it would also be possible to achieve an
efficient separating
equilibrium
without appealing
to convertible
bonds. For
example, suppose that good and bad firms behaved as before (i.e., financed with
long-term debt and equity, respectively), but that medium firms simply used
short-term debt to postpone an equity issue until time 1. This also satisfies the
conditions
for a separating equilibrium.’
But this strategy of simply postponing an equity issue works well only because
the current version of the model has the unnatural
feature that information
asymmetries
disappear completely after time 0. This feature implies that there
are no adverse-selection
problems associated with a time 1 equity issue. A more
complete model would have the property that there is a steady-state
level of
*For example, if costs of financial distress were very high. then medium and bad firms would
be unwilling ever to use long-term debt. Thus, in order for both types to invest. they would both
have to issue equity. This pooling equilibrium
cannot be sustained if N is small in relation to
(p - CJ)fX” - .Y,).
‘The logic is similar to that above. Since there is symmetric information
at time 1, the medium
firm knows it will be selling a fairly priced security. The bad firm will not mimic the medium strategy
ofdelaying
the equity issue, since if it deteriorates,
it will be unable to raise enough equity financing
to refinance its short-term debt, and will thus face costs of financial distress. Neither the bad nor the
medium firm will mimic the good firm by issuing long-term debt, again because of costs of financial
distress.
10
J.C. Srein. Concrrrible
bonds ax bockdoor
equiry financing
information asymmetry - when managers’ time 0 private information becomes
public at time 1, managers learn a new piece of private information. In this way,
managers are always one step ahead of the rest of the market, and the adverseselection problems associated with an equity issue cannot be avoided simply by
postponing the issue.‘O
it is straightforward to extend the model in this direction. Two new assumptions are needed: 1) At time 1, half of the medium firms (call them the Ma’s) get
some new private information that suggests they will do better than previously
expected - their probability of receiving X, rises from p to pG > p. The other
half of the medium firms (the MB’s) see their probability of receiving XH fall to
(2p - pG) c p. 2) At time 1, some of the firm’s existing assets can be liquidated
and the initial outlay of I recovered. But the NPV of this liquidation strategy is
negative, reducing the net value of the firm by L.
In the appendix, I prove that if (pc - p) and L are both large enough and the
NPV of the time 0 investment, N, is small enough, then there can no longer be
any separating equilibrium (efficient or otherwise) where the medium firms issue
short-term debt at time 0. The logic is as follows. If a medium firm receives
optimistic new private information at time 1 (i.e., becomes an Mo) it will then be
reluctant to go through with the planned equity issue at that time, perceiving
itself to be undervalued. This may result in an Mo instead inefficiently liquidating some assets to repay its short-term debt. I’ Folding back to time 0, there can
thus be e.x anre expected costs to a medium firm of relying on postponed equity
financing. We can therefore endogenously rule out the postponed equity strategy, leaving convertible bonds as the only means of achieving an efficient
separating equilibrium. lz
2.3, Discussion
The model is extremely stylized, and some of its specific conclusions are due to
oversimplified assumptions. For example, it is unrealistic to assume that a medium firm will be able to achieve conversion with certainty. At the expense of
a bit more notation, it would be possible to generalize the model so that
a medium firm is more likely to achieve conversion than a bad firm, but still
“Lucas
and McDonald
I 1990) present an infinite-horizon
steady-state
wherein managers always have one-step-ahead
knowledge of firm value.
“Such inefficient liquidation
at time 1 would
result that ‘good’ types may pass up positive-NPV
model of equity
be an exact analog to the Myers-Majluf
investments rather than issue underpriced
issues,
(1984)
equity.
“Central
to this result is the fact that an equity issue at time 1 is discretionary
- a firm can decide
to scrap a planned issue if its private information
is optimistic. If it were possible to contractually
precommit
to equity issues sufficiently far in advance. such precommitted
issues might represent
a viable means for overcoming
information
problems and restoring efficient investment.
J.C. Stein. Conrerlible
bonds as bockdoor equity financing
II
faces some uncertainty.
In this case, convertibles
would not wholly eliminate
expected costs of financial
distress. Nonetheless,
the qualitative
conclusion
- that convertibles
offer an attractive middle ground between the high expected
costs of distress associated with a debt issue and the large negative announcement impact associated with an equity issue - would remain unchanged.
In a richer model of this sort, however, it might be difficult to argue that
a convertible bond represents an optimally designed security. Although convertibles will continue to represent an improvement
over debt and equity for some
firms, one might imagine other financing contracts that now do even better
- contracts that completely eliminate costs of distress while at the same time
avoiding some of the informational
problems associated with a common equity
issue.
One example of such a contract can be given in the context of the model
presented above. Suppose that at time 0 the company issues a security that can
be redeemed at time 1 for k shares of stock, where the redemption
ratio
k depends on the stock price that prevails at time l- in particular, k is given by
l/P,. By construction,
this security is ‘adverse-selection-proof’.
No matter what
management’s
information
advantage at time 0, everybody can agree that the
security is worth $1. Furthermore,
since the security converts into equity with
certainty at time 1, there is no potential for costly financial distress.13
This line of reasoning suggests that the theory of convertible
bond issuance
presented here is only part of the story. Although the theory can explain why
some corporations
prefer convertibles
to straight debt or equity, it implicitly
limits the menu of financing options to these three instruments.
It would clearly
be desirable to have a model that endogenously
explains why the alternative
(and apparently
more efficient) kinds of securities described above are rarely
seen in practice.”
3. Empirical
implications
The theory of convertible
bond issuance presented above has a number of
empirical implications.
In this section, I discuss four categories of evidence that
are relevant for assessing the theory: 1) managers’ stated motivations
for using
“Brennan
(1986) discusses
exactly
such a security.
“Interestingly.
there have been several recent issues ofan instrument that resembles that sketched
above. For example, in May 1991, GM announced
plans to raise 5600 million by issuing a type of
‘preference’shares
(dubbed PERCS). One key feature of the PERCS is that they would automatically
convert into shares of GM common in July 1994. with the number of shares being a decreasing
function of the stock price prevailing at that time. Thus, as with the example given here. the PERCS
represent a form of financing that: 1) has a true value that is relatively insensitive to any inside
information
managers might have. yet 1) does not seem to involve any potential for costly financial
distress.
J.C. Stein. Concerrible bona2 as backdoor equityjinancing
I?.
convertibles, 2) characteristics of firms that rely heavily on convertibles, 3)
convertible call provisions and firm’s call policies, and 4) stock-price reactions
to announcements of convertible issues.
3.1. Managers’
motit’ations for using comertibles
Several researchers have used surveys to gather direct evidence on why
managers opt for convertible financing. An early such study is Pilcher (1955).
Pilcher’s survey asked (p. 60):
Which played the most important role in the decision by your company to
utilize the convertible privilege: the desire to ‘sweeten’ the senior leverage,
thereby making it more attractive to buyers, or the desire to raise common
equity on a sort of delayed action basis?
In this survey, 82% of the respondents chose the ‘delayed equity’ answer and
18% chose the ‘sweetened debt’ answer. Brigham (1966) asked an almost
identical question, and the pattern of responses was very similar - 73% indicated that their primary intent in issuing a convertible was to obtain equity
financing, whereas 27% indicated that they wished to sweeten a debt issue.
Brigham also asked those in the 73% majority why they used a convertible to
obtain equity financing. The overwhelming majority - 68% - said they believed
their stock price would rise over time, so that a convertible provided a way of
selling common stock at a price above the existing market.
Finally, Hoffmeister’s (1977) survey allowed respondents to choose from
among five other possible rationales besides delayed equity and sweetened debt.
The respondents were also asked to pick a first, second, and third choice. Again,
the delayed equity motive emerged as the single most important, ranking as first
choice for 34% of the firms and as one of the top three choices for 70%.15 In
sum, the overriding message from the surveys seems to be that many (if not
most) managers issue convertibles in the hopes of increasing the amount of
equity in their capital structures. Thus the survey data support the basic premise
of the model developed here.
In addition, the survey data help to underscore the important differences
between this model and that of Constantinides and Grundy (1989). In their
model, as in this one, the use of a convertible helps to overcome problems due to
asymmetric information - it allows for a separating equilibrium and efficient
investment. In the Constantinides-Grundy
model, however, the issue of a
15Hoffmeister also split his sample between industrial and financial firms, and found that the
delayed equity motive was disproportionately
important
for the industrial
firms. 17% of whom
ranked it as their first choice. In contrast,
sweetened debt was the most important
motive for
financial firms, with 41% picking it as their first choice.
J.C. Srein. Conrrrrrble bonds as backdoor equu.r financing
13
convertible must be combined with a publicly observed srock repurchase to have
the desired effect. ‘Bad’ firms are deterred from issuing convertibles not because
of costs of financial distress, but rather because they find it unattractive
to
repurchase overpriced stock.r6
This implication
of the Constantinides-Grundy
model is difficult to square
with the survey data - if managers say they are using convertibles
to get more
equity into their capital structures, they would certainly not want to use the
proceeds of a convertible issue to buy back stock. Other available evidence also
casts some doubt on the notion that convertible
issues are accompanied
by
stock repurchases.
For example, the papers of Dann and Mikkelson
(1984),
Eckbo (1986). and Mikkelson and Partch (1986) all contain data (taken from
issuer prospectuses)
on the planned use of proceeds from convertible
issues.
None makes any mention of share repurchases:
the most significant
uses of
proceeds are capital expenditures,
general corporate
spending, and debt refinancing.
In fact, debt refinancing is a very important
use of convertible proceeds. For
example, Dann and Mikkelson (1984, p. 175) note that ‘one-third of the issuances were virtually entirely refinancing of existing straight debt’. This is what
one might expect if (as the survey data suggest) managers use convertibles
in the
hopes of attaining a less-levered capital structure. It is, however, essentially the
reverse of what is predicted by the Constantinides-Grundy
model.
3.2. Characteristics
of concertible
issuers
The model of this paper suggests that convertibles
would be especially
valuable for firms that: 1) are characterized
by significant informational
asymmetries and 2) might incur large costs of financial distress if they added more
debt to their capital structures.
Broman (1963) presents some early evidence consistent with the latter prediction. He examines 60 industrial
firms that issued convertible
bonds of $10
million or more in the period 1951-1959, and finds higher leverage ratios for
these firms than for those issuing straight debt. Indeed, this finding leads
Broman to conclude that ‘debt advantages
[of convertibles]
are not as important in the minds of issuers as an eventual increase in equity ownership’ (p. 71).
More recently, Essig (1991) conducts a detailed study of the characteristics
of
convertible
issuers. Among the variables he examines are: 1) the ratio of R&D
to sales, 2) the ratio of tangible assets (property, plant, and equipment
plus
161n Constantinides
and Grundy (1989). as in the other theories of convertible bond issuance cited
above, little importance
is assigned to the callability feature. What is distinctive about a convertible
in their framework (i.e., what allows the combination
ofa convertible issue and a stock repurchase to
have the desired informational
properties) is simply that it is concave in firm value for low values of
the firm and convex for higher values.
II
J.C. Stein, Convertible bonds OSbackdoor equit~finuncing
inventories) to total assets, 3) the ratio of market value of equity to book value,
4) the ratio of long-term debt to equity, and 5) the standard deviation of changes
in cash Row. Using both simple stratifications of the data as well as a more
sophisticated multivariate regression framework, Essig relates these variables to
firms’ propensities to employ convertible financing.
A number of significant patterns emerge. I7 First, firms are more apt to rely on
convertibles if they have high ratios of R&D to sales. For example, firms that
make ‘heavy’ use of convertibles (i.e., their ratio of convertible debt to all debt
exceeds 67%) have R&D to sales ratios almost twice those of other firms. To the
extent that a high R&D ratio is indicative of the potential for information
asymmetries, this finding supports the prediction of the model.
Alternatively, one might argue that a high R&D ratio indicates that a firm has
important growth options, which - in the context of Myers (1977) - would tend
to make financial distress more costly. In other words, a high R&D ratio may be
an empirical proxy for a high value of the parameter c. Under this interpretation
too, the empirical relationship conforms to that predicted by the model.
Essig also finds that convertible use is strongly negatively related to the ratio
of tangible assets to total assets and strongly positively related to the marketto-book ratio. In both cases, one can interpret the results much like the results
for the R&D-to-sales ratio. For example, a low level of tangible assets might
make liquidation (and hence financial distress) costly. Similarly, a high marketto-book ratio would appear to indicate the presence of important growth
options, which again would make distress costly.
With regard to the ratio of long-term debt to equity, Essig’s results confirm
Broman’s (1963) earlier findings. Firms that have high debt-to-equity ratios are
significantly more likely to use convertibles. Finally, Essig documents that
convertible use is also positively linked to the volatility of a firm’s operating cash
flows. Again. these results are consistent with the model’s prediction that
convertibles should be particularly attractive to issuers facing potentially large
costs of financial distress.
3.3. Cull provisions nndjims’
cull policies
The model in this paper differs from other models of convertible bond
issuance in that it emphasizes the importance of the call provision. Convertible
bonds are typically callable after the expiration of a modest call protection
period - in Asquith’s (1991) sample of convertibles issued between 1980 and
1983, the median length of this call protection period is 252 days and 21% of all
convertibles have no call protection whatsoever.
“The patterns discussed below appear to be quite robust - they show up both in the simple
stratifications
and in a variety of multivariate
regression specifications.
J.C. Srein. Conrerrlble
bonds us buckdoor
equrt! finuncing
I5
Moreover.
Asquith documents
that, contrary
to widespread
belief, firms
actually do use the call feature to force prompt conversion after the expiration of
the call protection period. provided that this is feasible (i.e., the conversion value
exceeds the call price) and that there are no negative cash flow consequences
associated
with the call. For example, Asquith finds that for firms where
dividends are less than after-tax interest payments, conversion is forced almost
immediately
after the conversion
value reaches 120% of the call price - the
median delay from this point is only 18 days.
The result is that a large fraction of convertibles
wind up being converted into
equity a relatively short time after the initial issue date. In Asquith’s sample,
approximately
two-thirds of all convertible bonds issued (and not subsequently
removed from the sample because of merger) are eventually converted. Asquith’s
empirical findings are thus consistent with both the spirit of the story offered
here and with the expectations
of the managers in the surveys. They suggest that
firms issuing convertibles
have a good chance of seeing them turn into equity
financing within a reasonable
time.
3.4. Stock-price
reactions
to announcements
of comertihle
issues
The model developed
above also has implications
for the magnitudes
of
stock-price
reactions to financing announcements.
First and most directly, it
suggests that the announcement
of a convertible
bond issue should not be
interpreted as negatively by the market as the announcement
of an equity issue
of comparable
size. A convertible
issue reveals a firm to be of medium quality,
whereas an equity issue reveals a firm to be of bad quality.‘*
Table 1 summarizes
some relevant evidence, taken from several papers that
have studied these announcement
effects. The first column of the table demonstrates that the absolute magnitude of the effect is indeed larger for equity issues
than for convertible issues, by a factor of about two. And this simple comparison
may understate the true differences, because convertible issues tend to be larger
(as a fraction of firm value) than equity issues.
The second column of the table therefore divides the average announcement
impact (in percentage terms) reported in each study by the average issue size
(scaled by the market value of outstanding
equity). Now the impact of an equity
issue looks to be about three times as large as the impact of a convertible issue of
the same size - an equity issue has a negative impact on the order of 28%
‘*In the context of the model. the exact maznitude
of the stock-price
reaction to a convertible
issue depends on the prior distribution
of good. medium. and bad firms in the population.
For
example, if there are many good firms, a convertible issue that reveals a firm to be of medium quality
will be bad news. and will lead to a negative stock price reaction. Conversely. if there are few good
firms, a convertible
issue can have a positive impact on the stock price.
16
J.C. Stein.
Concertible
Table
Average
two-day
announcement
impact
equity finuncing
bonds us backdoor
1
for common
stock and convertible
Average
announcement
impact
Study
bond offerings.
Average
impact/issue
size
(A) Common stock ojhings
Asquith-Mullins
Masulis-Korwar
Mikkelson-Partch
Unweighted
(1986)”
( 1986)b
(1986)’
average
(B) Conrerrible
Dann-Mikkelson
Eckbo (1986)’
Mikkelson-Partch
Unweighted
- 31.0%
- 22.0%
- 29.5%
- 3.57%
- 27.5%
bond offrrings
(1984)”
(1986)’
average
“Source: Tables 2 and 3 (primary offerings
“Source: Average impact comes from table
table 3. column 2.
‘Source: Average impact comes from table
average issue size comes from table 3, row 2.
dSource: Average impact comes from table
‘Source: Average impact comes from table
- 3.0%
- 3.25%
- 4.46%
only).
5 (industrial
- 2.3 1%
- 1.25%
- 1.39%
- 10.5%
- 9.6%
- 6.2%
- 1.65%
- 8.8%
firms); average
issue size comes
4 (events with no contemporaneous
3; average
4; average
from
announcement);
issue size comes from table 2, column 2.
impact.issue
size is calculated on p. 149.
of issue size, while a convertible issue has a negative impact of roughly 9%
of issue size.
Of course, these comparisons need to be interpreted with caution, as they
completely overlook issues of sample selection. It would seem, however, that
controlling for issuer characteristics might well strengthen the contrast between
equity and convertible issues. Both the theory and much of the evidence
reviewed in section 3.2 above suggest that convertible issuers are exactly the
kinds of firms that would otherwise be expected to suffer from particularly large
announcement effects. as they are the ones for whom asymmetries of information are most pronounced. Thus our observation of smaller announcement
effects with convertibles than with straight equity is all the more striking.
The model may also shed light on certain cross-sectional aspects of stockprice reactions to convertible issues. For example, Mikkelson and Partch (1986)
document that convertibles with high bond ratings (A and above) have very
negative announcement effects. whereas convertibles with low ratings (B and
below) have essentially no announcement effects. (See their table 7.) Mikkelson
and Partch, as well as other authors [e.g., Brennan and Kraus (1987)], have
J.C. Stein, Camerrihle
bonds us backdoor equri>‘_financing
17
argued that this finding is difficult to reconcile with existing theory. But the
current model provides a simple explanation:
the greater the potential for costly
distress (i.e., the lower the bond rating) the more credible is the convertible
as
a signal of optimism. Firms with low bond ratings have the most to lose if they
are unable to force conversion, and hence will only issue convertibles if they are
quite optimistic about the prospects for their stock price.”
4. MCI’s use of convertibles, 1978-1983”
The case of MCI Communications
Corporation
provides a particularly sharp
illustration
of many of the ideas developed
above. MCI was organized
in
response to a change in FCC policy that allowed new companies to enter the
market for specialized long-distance
services. MCI went public in June 1972,
with an issue of $30 million of common stock. The company experienced large
operating losses in its first few years.
By 1978, however, the outlook had improved considerably,
in large part because of the success of MCI’s Execunet service. Execunet. which was introduced
in 1974, enabled MCI to attract small-business
subscribers who could not afford
dedicated private lines between particular cities. Unfortunately,
the growth of
Execunet was constrained
in its early years by a 1976 court order that restricted
it to existing customers.
The court order was lifted in May 1978. At that point MCI embarked on
a period of dramatic growth. Total assets went from $161 million in March 1978
to $2,071 million in March 1983. This growth implied a need for repeated large
infusions of external financing.
At the time this rapid growth began, MCI was highly leveraged. In March
1978, total debt stood at 5173 million. Thus in book-value terms, the company
had a ratio of debt to total capital in excess of 100%. Even in market-value
terms, the picture was not much better - the market value of common stock at
this time was only in the neighborhood
of S40 million. An application of Myers’s
“One should probably not place too much inferential weight on this particular empirical finding,
however. First, the sample is quite small. Second, an apparently
contradictory
result is reported by
Eckbo (1986). Finally. there are again issues of sample selection - for example, low-rated issuers are
probably subject to greater information
asymmetries,
and hence one might expect a larger price
impact. all else being equal.
“‘Most of the factual material in this section is drawn from the Harvard Business School case
study ‘MCI Communications
Corp., 1983’by Bruce Greenwald (19811. Moreover. in a teaching note
for the XlCI case, Greenwald
(1986, p. 8) makes a verbal argument for convertibles
that closely
parallels the formal one offered here: ‘(Convertibles)
offer a promise of ultimately escaping the
business-risk burdens of debt, while showing management’s
confidence in the future of the company.
If the price of the stock does not rise above the price at which conversion can be forced, management
must live with the burden of unconverted
debt.’ My debt to Greenwald’s analysis of the MCI case
should be apparent from this passage.
IS
J.C. Srrin. Conrrrriblr
bonds us buckdoor
eyuir~Jmmcing
(1977) logic suggests that, given the obvious importance of future investment in
its growth options at this point, excessive debt could have been particularly
damaging to a company like MCI.
MCI’s high debt level and the accompanying potential for costly financial
distress would seem to have dictated that the external funds be raised primarily
through some sort of equity instrument. But MCI management expressed an
aversion to the issuance of straight equity. Chief financial oficer Wayne English
was quoted as saying: ‘It was always our conviction that issuing more common
would knock the props out from under our stock.’
MCI decided to finance much of its growth with the use of convertible
securities. At first, convertible preferred stock was used. Convertible preferred
was apparently chosen over convertible debt because of tax considerations
- MCI was initially unable to take advantage of the deductibility of interest
expense because of a large accumulation of tax losses. A first issue of convertible
preferred in December 1978 raised a gross amount (before issue expenses) of 928
million. This was followed by a second offering in September 1979 that
raised S67.5 million, and a third offering in October 1980 that raised S49.5
million.’ 1
The call provision on these securities allowed MCI to call them any time the
market price of MCI common exceeded the conversion price by a stated margin
(e.g.. 25%) for 30 consecutive trading days. As its stock price rose, MCI was
therefore able to force prompt conversion of all three convertible preferred
issues, with the last conversion being effected in November 1981.
With its debt ratio improving, MCI undertook a $52.5 million sale of public
subordinated debentures in July 1980. Another bond issue in April 1981 raised
S105.9 million. The company then returned to the use of convertible securities,
although it switched from using convertible preferred to convertible subordinated debentures. An August 1981 convertible debt issue raised SlOO million
and a May 1982 issue raised another $250 million. Again, the use of early call
provisions combined with a rising stock price enabled MCI to force prompt
conversion of these two issues: they were both converted by February 1983.
Thus, in a little over four years MCI was able to force conversion on five
consecutive convertible issues, representing total financing of almost S500 million. A sixth convertible issue - another convertible debenture - in March 1983
produced an additional $400 million. Finally, in July 1983, MCI raised a record
St billion with a ‘synthetic’ convertible, consisting of a package of bonds and
detachable callable warrants.
“The theoretical
arguments
made above apply to convertible
preferred to the extent that an
excess of preferred in a company’s capital structure can lead to costs of financial distress. Although
preferred may not have exactly the same properties as debt in this regard. its seniority relative to
common equity implies that it can nonetheless create the kinds of ‘debt overhang’ costs identified by
Myers (1977).
Unfortunately
for MCI, it was unable to force the conversion
of these
last two issues. Its stock price, which was in the low $40~ at the time of
both issues, began a sharp decline as MCI fared poorly in product market
competition
with AT&T. This left MCI with a large debt burden. which it had
difficulty servicing. In apparently
desperate circumstances,
MCI sold an 18%
equity stake to IBM in June of 1986. at a price of approximately
914 per share.
In December 1986, it announced
major layoffs and large reductions in capital
expenditures.
Although
the successful conversions
of the first five issues illustrate how
convertibles can be used as an indirect method of obtaining equity financing, the
failures of the last two underscore that this method is not without its risks. As
seen in the model of section 2, it is exactly the presence of these risks - the
potential for costly distress when stock prices fall and conversion
cannot be
forced - that enable convertibles
to be issued with less adverse informational
consequences
than straight equity.
5. Summary
This paper has argued that companies may find convertible bonds an attractive middle ground between the negative informational
consequences
associated
with an equity issue and the potential for costly financial distress associated with
a debt issue. When used with a call provision that enables early forced conversion, a convertible
can serve as an indirect (albeit somewhat risky) mechanism
for implementing
equity financing that entails less of an adverse price impact
than an offering of common stock.
The theory developed
here has a number
of empirical
implications
that
appear to be supported by existing evidence. In particular, the theory fits well
with the following facts: 1) a convertible
issue typically leads to a less negative
announcement
effect than does an equity issue of comparable
size; 2) convertibles tend to be used predominantly
by highly-leveraged,
highly-volatile
firms
with large R&D portfolios and above-average
levels of intangible
assets; 3) a
majority of managers assert that their primary motive in using convertibles
is to
raise equity on a ‘delayed action’ basis; and 4) most convertible
issues are
forcibly converted promptly after the expiration of their call protection periods,
provided such forced conversion
is possible and does not have negative cash
flow consequences.
Appendix
It remains to be shown that, for (po - p) and L sufficiently
N sufficiently
small, there cannot
be any separating
equilibrium
medium firms issue short-term
debt at time 0.
large and
in which
Suppose that a medium firm did issue short-term debt at time 0. Let us first
ask whether we can have a pooling equilibrium
in the time 1 subgame, so that
both dCfo and MB types issue equity at that time - in other words, can
we support the delayed equity strategy’? The pooling outcome requires that
an MC prefer issuing equity at the pooling price to not issuing and liquidating
some of its assets instead. This implies that there can be no pooling
if
I( p. - p)(X, - XL)/(pXH + (1 - p)X,) > L. Thus if (po - p) is sufficiently
large. there will be no pooling at time 1.
This means that one of two things must happen at time 1 - either the Mo
types will liquidate at cost L or they will refinance with debt and incur expected
bankruptcy
costs of (1 - po)c. So, working backword, the e.x anre expected
profits to a medium firm in the posited separating equilibrium
cannot exceed the
maximum of (N - f.12) and (N - (1 - po)c/2). Thus, if N is small enough in
relation to L and c, e.~ ante expected profits will be negative. This destroys the
posited separating
equilibrium,
as a medium firm would be made better off
simply by doing nothing - not issuing and not investing - at time 0.
References
Asquith. Paul. 1991. Convertible debt: A dynamic test of call policy, Working paper (Sloan School of
Management.
MIT. Cambridge.
MA).
Asquith. Paul and David Mullins. 1986. Equity issues and offering dilution, Journal of Financial
Economics
15, 61-89.
Asquith. Paul and David Mullins. 1991. Convertible
debt: Corporate
call policy and voluntary
conversion, Journal of Finance 46. 1273-1289.
Brennan. Michael, 1986. Costless financing policies under asymmetric information,
Working paper
(University of British Columbia,
Vancouver.
BC).
Brennan, Michael and Alan Kraus. 1987. Efficient financing under asymmetric information,
Journal
of Finance 42, I X5- 1243.
Brennan. Michael and Eduardo Schwartz. 1977. Convertible
bonds: Valuation and optimal strategies for call and conversion. Journal of Finance 32. 1699-1715.
Brennan. Michael and Eduardo Schwartz. 1980. Analyzing convertible bonds, Journal of Financial
and Quantitative
Analysis 15, 907-929.
Brennan, Michael and Eduardo
Schwartz.
1985. The case for convertibles.
Journal
of Applied
Corporate
Finance I. 55-6-t.
Brigham, Eugene, 1966, An analysis of convertible debentures: Theory and some empirical evidence,
Journal of Finance 21. 35-54.
Broman. Keith, 1963, The use of convertible subordinated
debentures by industrial firms 1949-59,
Quarterly
Review of Economics and Business 3, 65-75.
Constantinides.
George and Bruce Grundy, 1989. Optimal investment with stock repurchase
and
tinancing as signals. Review of Financial Studies 2. 445-465.
Dann. Larry and Wayne Mikkelson, 198-t. Convertible debt issuance, capital structure change and
financing-related
information.
Journal of Financial Economics
13, 157-186.
Diamond. Douglas. 1984, Financial intermediation
and delegated monitoring,
Review of Economic
Studies 51, 393-414.
Eckbo, B. Epsen. 1986. Valuation effects of corporate debt offerings, Journal of Financial Economics
15, 119-151.
Essig. Stuart. 1991, Convertible
securities and capital structure determinants.
Ph.D. dissertation
(Graduate
School of Business. University of Chicago, Chicago, IL).
J.C. Srem. Concerrible
bonds us backdoor
equi[yJinnncing
21
Green. Richard, 1984, Investment incentives, debt and warrants. Journal of Financial Economics 13.
115-136.
Greenwald. Bruce. 1984, MCI Communications
Corp.. 1983. Harvard Business School case services
no. 284-057 (Harvard
Business School, Boston. MA).
Greenwald.
Bruce, 1986, MCI Communications
Corp.. 1983 - Teaching note, Harvard Business
School case services no. 386-l 10 (Harvard Business School. Boston, MA).
Harris, Milton and Artur Raviv, 1985. A sequential signalling model of convertible debt call policy,
Journal of Finance 41, 815-830.
Hoffmeister, J. Ronald. 1977. Use of convertible debt in the early 1970s: A reevaluation
of corporate
motives, Quarterly
Review of Economics and Business 17. 23-32.
Ingersoll. Jonathan.
1977a. A contingent
claims valuation
of convertible
securities, Journal of
Financial Economics 4. 289-322.
Ingersoll. Jonathan,
1977b. An examination
of corporate
call policy on convertible
securities.
Journal of Finance 32. 463-478.
Jaffee. Dwight and Andrei Shleifer, 1990, Costs of financial distress, delayed calls of convertible
bonds. and the role of investment banks, Journal of Business 63, S107-S124.
Lucas, Deborah and Robert McDonald.
1990, Equity issues and stock and price dynamics. Journal
of Finance 45, 1019-1043.
Masulis. Ronald and Ashok Korwar. 1986, Seasoned equity offerings: An empirical investigation,
Journal of Financial Economics
15, 91-l 18.
Mikkelson. Wayne, 198 I, Convertible calls and security returns. Journal of Financial Economics 9.
237-264.
Mikkelson, Wayne and Megan Partch, 1986. Valuation effects of security offerings and the issuance
process. Journal of Financial Economics
15, 31-60.
Myers, Stewart, 1977, Determinants
of corporate
borrowing,
Journal of Financial Economics 5.
147-175.
Myers, Stewart and Nicholas Majluf. 1984, Corporate
financing and investment decisions when
firms have information
that investors do not have, Journal of Financial Economics 13. 187-221.
Pitcher, C. James. 1955, Raising capital with convertible securities. Michigan business studies no.
21.2 (University
of Michigan, Ann Arbor, MI).